A function has an isolated singularity at if is analytic on a punctured disk but not at itself — either is undefined at or fails to be analytic there.

Isolated means: there is a neighborhood of in which is the only point where misbehaves. Functions like near have non-isolated singularities; those are rarer and need different tools.

Three types

Distinguished by the principal part of the Laurent series at :

Removable singularity

The Laurent series has no negative terms — it’s actually a Taylor series. Equivalently:

  • exists and is finite.
  • is bounded near .

The singularity can be “removed” by defining , making analytic at .

Test: exists and is finite. Or: is bounded near .

Example. at . Laurent / Taylor: — no negative terms. Limit is . Removable.

Pole of order

The Laurent series has finitely many negative terms, with the most negative being , . Equivalently:

  • exists and is nonzero.

Test: blows up like at .

Examples.

  • has a simple pole (order 1) at .
  • has a pole of order 3.
  • has a simple pole at and a pole of order 2 at .
  • has a simple pole at (since is analytic and nonzero at ).

Essential singularity

The Laurent series has infinitely many negative terms. The function does crazy things near .

Picard’s theorem (deeper): in every neighborhood of an essential singularity, takes every complex value with at most one exception, infinitely many times.

Examples. , , — all at .

For most practical problems, essential singularities don’t come up. Poles and removable singularities cover the standard cases.

Recognizing the type quickly

  • If has a known formula and you can compute the limit at : limit exists removable; limit is infinite pole or essential.
  • For pole vs. essential: multiply by for increasing and check if the limit becomes finite (pole) or never finite (essential).
  • Quick: a rational function in lowest terms has poles at zeros of , of order equal to the multiplicity of that zero in . Nothing essential.
  • Functions involving , , etc., where have essential singularities there.

Residues

The behavior near a singularity is encoded in the residue — the coefficient of in the Laurent series. Different singularity types have different formulas:

  • Removable: residue is .
  • Simple pole: .
  • Pole of order : .
  • Essential: read off the Laurent expansion directly; no formula.

See Residue (complex analysis) for worked computations.

In context

The classification matters because:

  • The Residue theorem requires evaluating residues; different types need different formulas.
  • Removable singularities contribute zero to closed-contour integrals.
  • Poles contribute each.
  • Essential singularities also contribute , but the residue must be extracted from the Laurent series by hand.